Registered Chinese name上海灵初智能科技有限公司Shanghai PsiBot Intelligent Technology Co., Ltd.
Founded in 2024, Lingchu Intelligence is a cutting-edge technology company focused on overcoming the problems of dexterous operation of embodied intelligence. The company brings to…
Ψ-SynRobotWorld Model converts human data and robot data
Founded in 2024, Lingchu Intelligence is a cutting-edge technology company focused on overcoming the problems of dexterous operation of embodied intelligence. The company brings together top talents from the robotics industry and top universities, and builds a joint laboratory with Peking University to build a software tool chain with the end-to-end PSI series model as the core. The Psi-SynEngine embodied data acquisition engine independently developed by Lingchu Intelligence can obtain massive high-quality real scene data at a very low cost. It has completed small-scale scene verification in real logistics customer warehouses and achieved a substantial breakthrough in sorting efficiency.
At the beginning of its establishment, the company received investment from well-known institutions such as Hillhouse Venture Capital and Lanchi Venture Capital. In March 2026, it announced the completion of angel round and Pre-A round financing of a total of 2 billion yuan. It has received strong support from many "national teams" and industrial capitals such as CDB Financial, SDIC Pioneer, Xuhui Capital, etc., and has become a core innovation force under the national embodied intelligence strategic layout.
02
MAIN PRODUCTS
Key Products
02 ITEMS
01
Ψ-SynRobot
ψ-SynRobot is a wheeled, dual-arm universal embodied robot independently developed by Lingchu Intelligence. It is equipped with the latest PSI base model and is specially built for complex logistics, retail and industrial scenarios.
Long-range autonomous operation: supports continuous tasks for more than 30 minutes, and has the ability to autonomously recover after task interruption
Dexterous operation: joint force control enables safer and more precise operations, with an accuracy of 99.9%
Multi-scenario skills: grabbing, placing, carrying, loading and unloading fixtures, scanning codes, labeling, sorting, using packages and tools (code scanning guns, magnetic pens, vacuum balls, etc.)
02
World Model converts human data and robot data
Lingchu's self-developed world model demonstrates cross-ontology migration capabilities, allowing the world model to serve as a bridge for robots to understand the physical world. It directly converts human data into robot data through the world model, and uses the converted data for model training.
When people use Lingchu data mining equipment to perform random operations, the model can understand the actions collected by the data mining equipment, complete the mapping of action space and image space, and ultimately efficiently present the conversion of human data to robot training data.
